commit e82793520f05609fad29c654a09d17971d0fb07f Merge: 78a8b93 43bc1e7 Author: Vladimir Sokolovsky Date: Wed Sep 28 22:16:36 2011 +0300 Merge branch 'ofed_kernel' of git://git.openfabrics.org/~swise/ofed_kernel into ofed_kernel_1_5 commit 78a8b93e427950bdea494ecfa2613f9f75e7495c Author: Faisal Latif Date: Tue Sep 27 16:43:39 2011 -0500 RDMA/nes: Backport fixes for PAU Backport fix for the Packed and Unaligned (PAU) code. Signed-off-by: Tatyana Nikolova Signed-off-by: Faisal Latif commit 43bc1e73f2357f38b0b9c73317c0eb7913f2472f Author: Steve Wise Date: Wed Sep 28 10:05:20 2011 -0500 RDSRDMA: Bring OFED up to date with upstream kernel Pull in commits 52fc50940e73bbb38cf12387db3495664cd57ba4 and 34b04e253ba8102ec828271565e4c218be4076c8 in OFED. These commits fix the PAGE_MASK interpretation and segment unmapping issues in RDSRDMA Signed-off-by: Jonathan Lallinger Signed-off-by: Steve Wise commit d80b23789ffb1b90ef42e285ca212cc9f7c835a4 Merge: ef492ff faa71aa Author: Vladimir Sokolovsky Date: Tue Sep 27 09:54:24 2011 +0300 Merge branch 'ofed_1_5_gireesh' of git://git.openfabrics.org/~vipulp/ofed_kernel into ofed_kernel_1_5 commit ef492ffb72250e0c457fb1b97470e5f0cd6aa61c Author: Mitko Haralanov Date: Mon Sep 26 17:14:02 2011 -0400 QLogic: Add support for HP BL685 blades Add SERDES tunning support for the HP BL685 full-height blades which have 3 Mez slots. Signed-off-by: Mitko Haralanov Signed-off-by: Mike Marciniszyn commit 956daab1693ce7d0cc39139d057f6d2e8cf260d3 Author: Mitko Haralanov Date: Mon Sep 26 13:02:18 2011 -0400 truescale.cmds: enhance CPU assignment truescale.cmds is enhanced to make the following changes to CPU assignments: - use the affinity_hint file if present, otherwise changes as below - non-kctx interrupts are assigned to the HCA's first local CPU - kctx interrupts are assigned from the second with a potential wrap back to the second Signed-off-by: Mike Marciniszyn commit 76094c8ce83af2d8f85d50d7c572a823ff6cc13c Merge: 7957aae 0e805f1 Author: Vladimir Sokolovsky Date: Mon Sep 26 16:53:02 2011 +0300 Merge branch 'ofed_1_5' of git://git.openfabrics.org/~vipulp/ofed_kernel into ofed_kernel_1_5 commit 0e805f17e51bb1d74cb2c60b1e98e9731f8128c0 Author: Kumar Sanghvi Date: Thu Aug 18 18:06:23 2011 +0530 MPA_v2: Backports for iw_cxgb4, iw_cxgb3 and amso Adds MPA_v2 backports to support Enhanced RDMA Negotiation. For iw_cxgb3 and amso, currently max supported ird/ord values are passed upwards. Adds support to following kernels: -2.6.39 -2.6.38 -2.6.37 -2.6.36 -2.6.35 -2.6.34 -2.6.33 -2.6.32 -2.6.32-EL6 -2.6.32-EL6.1 -2.6.32_sles11sp1 -2.6.31 -2.6.30 -2.6.18-EL5.7 -2.6.18-EL5.6 -2.6.16_sles10sp4 Signed-off-by: Kumar Sanghvi commit b63fc7fe9a5406e0aed55aa2854fc3d2f96236dd Author: Kumar Sanghvi Date: Thu Aug 18 03:49:42 2011 +0530 IB/core: Propagate ird/ord values upwards Updates iw_cm_event to support propagating the ird/ord values upwards to the application Signed-off-by: Kumar Sanghvi commit c1078277b1112b719925f44e4815c50d5086e91a Author: Kumar Sanghvi Date: Mon Aug 8 16:28:51 2011 +0530 iw_cxgb4: Add support for MPA V2 Enhanced RDMA Negotiation The patch adds support for Enhanced RDMA Connection Establishment (draft-ietf-storm-mpa-peer-connect-06). Details of draft can be obtained from: http://www.ietf.org/id/draft-ietf-storm-mpa-peer-connect-06.txt The patch updates the following functions for initiator perspective: -send_mpa_request -process_mpa_reply -post_terminate for TERM error codes -destroy_qp for TERM related change -adds layer/etype/ecode to c4iw_qp_attrs for sending with TERM -peer_abort for retrying connection attempt with MPA_v1 message -added c4iw_reconnect function The patch updates the following functions for responder perspective: -process_mpa_request -send_mpa_reply -c4iw_accept_cr -passes ird/ord to upper layers Signed-off-by: Kumar Sanghvi commit 7957aae129c300d0b5cff8528fba00af8cce8cc3 Author: Faisal Latif Date: Sun Sep 25 21:19:06 2011 -0500 RDMA/nes: Support for Packed And Unaligned fpdus Support for PAU is needed for interoperability between NES and other non-NES nodes. When the NES hardware detects a PAU frame, it will pass it to the driver to process the frame. NES driver creates a new frame for each FPDU and forwards it to the hardware to be sent to its associated qp. Signed-off-by: Tatyana Nikolova Signed-off-by: Faisal Latif commit 729040de6e583e7b1ae1f7f0be808d3e254b9e56 Author: Faisal Latif Date: Sun Sep 25 21:18:43 2011 -0500 RDMA/nes: Support for MPA revision 2 MPA V2 changes are based on an IETF STORM Draft, which will update RFC5043 and RFC5044. MPA V2 extends the configuration negotiation for RDMA connection establishment. For backwards compatibility, MPA V2 enabled driver reverts to MPA V1, if the remote node doesn't support MPA V2. Signed-off-by: Tatyana Nikolova Signed-off-by: Faisal Latif commit cd041554521ce46cb443883ceddaa515e71c3736 Author: Faisal Latif Date: Sun Sep 25 21:18:31 2011 -0500 RDMA/nes: Print ip address for critical errors Prints the IP address of the remote host when critical asynchronous event is received. Signed-off-by: Tatyana Nikolova Signed-off-by: Faisal Latif commit cf8cdcc1e6fe328caf3170349409c696f575ebd6 Author: Faisal Latif Date: Sun Sep 25 21:15:56 2011 -0500 RDMA/nes: Fix terminate connection Fixes a crash that occurs during close when error async event is received. Terminate message is not sent to the remote node if already processing close. Signed-off-by: Tatyana Nikolova Signed-off-by: Faisal Latif commit faa71aa633e1e6aa96070d51d03e84a57b29ea4f Author: Gireesh Nagabhushana Date: Mon Sep 26 17:35:59 2011 +0530 cxgb4: change version to 1.3.0-ofed to differentiate between base and ofed. Signed-off-by: Gireesh Nagabhushana commit a7f7c501f9cf0a1454323c5f6f8668be2ddf2b1d Author: Mike Marciniszyn Date: Fri Sep 23 12:04:24 2011 -0400 IB/qib: Add logic for affinity hint Call irq_set_affinity_hint to provide user mode programs like irqbalance the information to be able to distribute qib interrupts appropriately. The logic allocates all non-receive interrupts to the first CPU local to the HCA. Receive interrupts are allocated round robin starting with the second CPU local to the HCA with potential wrap back to the second CPU. Signed-off-by: Mike Marciniszyn commit 96ecdac8d78e2d20b32eb4ec50643bcb1c03f828 Author: Mike Marciniszyn Date: Fri Sep 23 11:10:49 2011 -0400 IB/qib: Add irq name refinements This patch adds a refinement to the name registered with MSIX interrupts so that user level scripts can determine the device associated with the IRQs when there are multiple HCAs with a potentially different set of local CPUs. Signed-off-by: Mike Marciniszyn commit d3d7ea0a142f1545f9bfbcd9e3bd47b7b2f9ce79 Author: Mike Marciniszyn Date: Fri Sep 23 10:59:03 2011 -0400 IB/qib: remove s_lock around header validation Observation in qib_ruc_check_hdr() shows that the s_lock is not required in the normal case. The r_lock is held in all cases, and protects the qp fields that are read. The s_lock will be needed to around the call to qib_migrate_qp() to insure that the send engine sees a consistent set of fields. Signed-off-by: Mike Marciniszyn commit 6aeceea336753dfc6b2987594b228e65b3a46982 Author: Mike Marciniszyn Date: Fri Sep 23 10:46:13 2011 -0400 IB/qib: memcpy optimizations The default memcpy used by qib_copy_sge() ends up being a rep movsb on x86_64, which is pretty slow. This fix adds an x86_64 specific routine that 1) probes for X86_FEATURE_REP_GOOD and 2) uses an inline asm routine builton rep movsq that testing has shown is better than the builtin memcpy for all cases up to 4K. The probing routine is now called when the qib module is loaded to enable the optimization. When X86_FEATURE_REP_GOOD is not set, the routine uses the kernel's unrolled __memcpy when the length is more than 64 and the builtin memcpy otherwise. This patch also adds the cache bypass copies from older releases. Testing has shown that AMD cpus benefit with a 40% improvement in netperf/ipoib. The cache_bypass_copy module parameter can be used to enable on non-AMD CPUs. The qib_verbs_send_dma() and qib_copy_from_sge are also changed to use the memcpy_string_op() to improve packet delivery performance to the send engine. The existing copy as well as a new stub probe routine are maintained as weak symbols for other architectures. Signed-off-by: Mike Marciniszyn commit 9ecf3abd1d880255e2f9be0e9714dcd49c97fede Author: Mike Marciniszyn Date: Fri Sep 23 09:44:07 2011 -0400 IB/qib: precompute timeout jiffies to optimize latency A new field is added to qib_qp called timeout_jiffies. It is initialized upon create and modify. The field is now used vs. the computation based on qp->timeout. Signed-off-by: Mike Marciniszyn commit b718e7f17cc9fc9d31148ec046bf071bdd2c105d Author: Mike Marciniszyn Date: Thu Sep 22 14:43:48 2011 -0400 IB/qib: qpn lookup optimizations The heavy weight spinlock in qib_lookup_qpn() is replaced with the RCU locking mechanism. The hash list itself is now accessed via jhash functions vs. the mod. The changes should benefit multiple receive contexts in different processors by not contending for the lock to just read the hash structures. The patch also adds a lookaside_qp (pointer) and a lookaside_qpn in the context. The interrupt handler will test the current packet's qpn against lookaside_qpn if the lookaside_qp pointer is non-null. The pointer is NULL'ed when the interrupt handler exits. Signed-off-by: Mike Marciniszyn commit dd2e48e7c02c0925a5e603d1a9752d20b7e15ed3 Author: Mike Marciniszyn Date: Thu Sep 22 11:50:04 2011 -0400 IB/qib: Eliminate divide/mod in converting idx to egr buf pointer The context init now saves a shift from rcvegrbufs_perchunk rcvegrbufs_perchunk_shift using ilog2. A BUG_ON protects the power of 2 assumption. Signed-off-by: Mike Marciniszyn commit 3eef5499db4d04aae5d10580b3735696957cd99d Author: Mike Marciniszyn Date: Thu Sep 22 11:40:13 2011 -0400 IB/qib: decode path mtu optimization Store both the encoded and decoded mtu in the qp structure as a minor optimization UC/RC receive routines. Signed-off-by: Mike Marciniszyn commit 2871fc8a45df95c52cac9707e886383ec712a20e Author: Mike Marciniszyn Date: Thu Sep 22 11:24:39 2011 -0400 IB/qib: Optimize RC/UC code by IB operation The memset for zeroing the work completion had been unconditional. This patch removes the memset and moves the zeroing into the work completion with a more explicit field by field set. With this patch, non-ONLY/non-LAST packets will avoid the overhead since they will not generate a completion. Signed-off-by: Mike Marciniszyn commit 828a7f6b95f5c29d3ddc93c8a5e01a83ccffb140 Author: Eli Cohen Date: Thu Sep 22 14:11:33 2011 +0300 ib_core: Change link_layer string to match mainline kernel In the mailinline it appears for IB as "InfiniBand" while in ofed it was "IB". Change to "InfiniBand". Signed-off-by: Eli Cohen commit 3f109aa8414c9d9ae0267c38bec41b597c759ddb Merge: 9f978fa 6d13d18 Author: Vladimir Sokolovsky Date: Thu Sep 22 09:12:47 2011 +0300 Merge branch 'ofed_kernel_1_5' of git://git.openfabrics.org/~mmarciniszyn/linux-2.6.to_ofed into ofed_kernel_1_5 commit 9f978fa3ef3d4826fa36b529d8f216a5bbf21cff Author: Shawn Bohrer Date: Wed Sep 21 08:26:51 2011 -0500 Fedora 15 is calling 3.0 2.6.40 so match that for backports. Fedora 15 has released a 3.0 kernel but is calling it 2.6.40 to avoid compatability issues with components expecting a 2.6.* kernel. Signed-off-by: Shawn Bohrer commit 6d13d18acfea76fbfa823b4c7f3bf57a1da2f39e Author: Mike Marciniszyn Date: Wed Sep 21 12:31:12 2011 -0400 IB/qib: Correct nfreectxts for multiple HCAs The code that was recently introduced to report the number of free contexts is flawed for multiple HCAs: /* Return the number of free user ports (contexts) available. */ return scnprintf(buf, PAGE_SIZE, "%u\n", dd->cfgctxts - dd->first_user_ctxt - (u32)qib_stats.sps_ctxts); The qib_stats is global to the module, not per HCA, so the code is broken for multiple HCAs. This patch adds a qib_devdata field, freectxts, that reflects the free contexts for this HCA. Signed-off-by: Mike Marciniszyn Reviewed-by: Ram Vepa commit 0c18cc2c5e7d18d637c8b7a6a789976fc25ad947 Author: Vipul Pandya Date: Mon Sep 19 18:51:30 2011 +0530 RDMA/cxgb3: Bring upstream commit b955150ea784af4c193b708a2e8091673bf23004 When a user QP is marked in error, also mark the CQs in error The flushing of work requests for user QPs is implemented entirely in the user mode library. The only kernel interaction is to mark the user QP object indicating it is in error when the QP exits RTS. When the user QP operations are called by the application (eg: post_send, post_recv), the QP in error bit is checked and if set, the library flushes the QP. If, however, the application is not doing IO, but rather just polling the CQ, it will never get flushed work requests. This breaks some classes of applications. This patch adds logic to mark user CQs in error when a QP that is bound to the CQ is marked in error. The library poll code can then notice the CQ is in error and flush all the in error QPs bound to that CQ. Design: - add 1 extra CQE entry to the CQ memory that will be used to indicate in error status. - return the desired CQ memory size that should be mapped by the library - bump the ABI since the create_cq uverbs response changes. - detect older libraries and reduce the mmap size accordingly. (The ABI bump doesn't break old libraries, since they didn't check the ABI field anyway) Signed-off-by: Vipul Pandya commit 4c6a5a907d47c1e10f54cf3ff79582c876e6ad84 Merge: a3468d0 8676af5 Author: Vladimir Sokolovsky Date: Wed Sep 21 07:53:41 2011 +0300 Merge branch 'ofed_kernel_1_5' of git://openfabrics.org/~moranp/ofed_1_5 into ofed_kernel_1_5 commit a3468d018e949499b6d656d90e0dafbd34b6b94c Author: Jeff Becker Date: Tue Sep 20 23:26:39 2011 +0300 NFSRDMA fixes for SLES11 SP1 Signed-off-by: Jeff Becker commit 8676af5ef82651fd9d48d2d3160cb20d788ca9b3 Author: Moran Perets Date: Sun Sep 4 11:55:45 2011 +0300 sdp: Fix Bug 2283 - Multi connection net_perf causes server to hang Fix the soft lockup bug by changing the allocation flag in sdp_bcopy.c from 0 to gfp. Reviewed by: Amir Vadai Signed-off-by: Moran Perets commit 501a88d202b9cd962b7473dfbf69332a0a8c25bd Author: Moran Perets Date: Mon Aug 1 16:34:19 2011 +0300 sdp: Fix backport sdp_0090_revert_to_2_6_24.patch Fix backport sdp_0090_revert_to_2_6_24.patch: sk_stream_rmem_schedule(sk, int) => sk_stream_rmem_schedule(sk, skb) Signed-off-by: Moran Perets commit f273be716d08aa1c8df081f477bb17901cd44947 Author: Moran Perets Date: Sun Jul 24 16:12:07 2011 +0300 sdp: Add "#include_next " to kernel_addons /net/sock.h Added "#include_next " to kernel_addons /net/sock.h in kernels 2.6.22 to fix compilation error Signed-off-by: Moran Perets commit 18de1beb0952781459e41fdd0d4cfd6232656817 Author: Moran Perets Date: Thu Jul 21 11:35:16 2011 +0300 sdp: Add "#include_next " to kernel_addons /net/sock.h Added "#include_next " to kernel_addons /net/sock.h in kernels 2.6.20/21/22_slert10_sp2/23/24 to fix compilation error Signed-off-by: Moran Perets commit dc95af16b0aa1cdfc73f63274f5b800088865bcd Author: Vladimir Sokolovsky Date: Sun Jun 26 18:52:37 2011 +0300 ib_sdp.conf: load ib_sdp when using AF_INET6_SDP address family The ib_sdp module is loaded in two cases: 1. In /etc/infiniband/openib.conf: “LOAD_SDP=yes” 2. AF_INET_SDP address family socket creation is called. The AF_INET_SDP cause the installing of the ib_sdp because in file /ofed_scripts/ib_sdp.conf we have: alias net-pf-27 ib_sdp So, the issue is that on AF_INET6_SDP socket creation the ib_sdp module is not created. This patch add net-pf-28 to ib_sdp.conf Signed-off-by: Vladimir Sokolovsky commit 54744953b6e19d642b92c1aed2a871eb0ebc7248 Author: Amir Vadai Date: Tue Apr 26 17:02:23 2011 +0300 sdp: Fix backports Signed-off-by: Amir Vadai commit 0eb29400267d2ddeb6946068c9c2ec84a076c9b8 Author: Amir Vadai Date: Tue Apr 26 12:38:03 2011 +0300 sdp: RDMA read timeout is too big - set it to 5 secs Signed-off-by: Amir Vadai commit 82c1d5c8ed34a1f77abcb7d01ac8f172340ffe8e Author: Amir Vadai Date: Tue Apr 26 12:35:59 2011 +0300 sdp: dump traffic to sdpprf only on demand When setting 2nd bit of sdp_data_debug Signed-off-by: Amir Vadai commit 185d7ee5b767157a9c8c74fefe4d528526b59eff Author: Amir Vadai Date: Tue Apr 26 12:31:42 2011 +0300 sdp: fix backports Signed-off-by: Amir Vadai commit cc1c2131541a1ad270d7370a84a1e7f5fa465ae0 Author: Amir Vadai Date: Thu Apr 21 13:55:55 2011 +0300 sdp: changed memory accounting warning into debug Signed-off-by: Amir Vadai commit 46a1f4a7cef36d1ed2b057579e521a2b5651a1ad Author: Amir Vadai Date: Thu Apr 14 14:27:43 2011 +0300 sdp: Fix issues in sdpprf Signed-off-by: Amir Vadai commit ac7920be63880abdfd24794eb385d3abae15a6ff Author: Amir Vadai Date: Tue Apr 12 17:40:07 2011 +0300 sdp: Remove protection before sleep on RX No need for this protection, it was needed because of a bug previously fixed. Credit update could be sent even when credit reaches 2. Signed-off-by: Amir Vadai commit d4f584c527b5e9cdfc6fbda4f4f87464b1b38711 Author: Amir Vadai Date: Tue Apr 5 10:58:56 2011 +0300 sdp: Fixed some coverity issues Signed-off-by: Amir Vadai commit 0aa494046a76188cff182c0707473bc24845aece Author: Amir Vadai Date: Thu Mar 24 10:46:06 2011 +0200 sdp: Fixed compilation error on 2.6.18 RH5.5 Signed-off-by: Amir Vadai commit 2932a78033503cc64c9408dc975b6b24fe69bea2 Author: Amir Vadai Date: Tue Mar 22 16:20:34 2011 +0200 sdp: fix memory leak. sockets_allocated wasn't freed Signed-off-by: Amir Vadai commit b650e167d8d1418a5c80e8101e3a65bc7f1a3be4 Author: Amir Vadai Date: Sun Mar 20 18:37:16 2011 +0200 sdp: Removed spaces and tabs at end of lines Signed-off-by: Amir Vadai commit ca88c554ad95dfb2256aebd41f6497be7974a3e4 Author: Amir Vadai Date: Sun Mar 20 18:32:23 2011 +0200 sdp: fix sdpprf Signed-off-by: Amir Vadai commit e0d7ffb18bbf6a78835a7aeab805a058040db293 Author: Amir Vadai Date: Sun Mar 20 17:25:02 2011 +0200 sdp: Fixed bcopy statistics Signed-off-by: Amir Vadai commit 368b9ce64a30a267db0d1527f3e47cf49e459cee Author: Amir Vadai Date: Sun Mar 20 15:34:33 2011 +0200 sdp: Bad behaviour when setting low rcvbuf size Signed-off-by: Amir Vadai commit ed219923b83bc454ef6897e16f3ff76225fff659 Author: Amir Vadai Date: Wed Mar 9 10:35:15 2011 +0200 sdp: Fixed a typo Signed-off-by: Amir Vadai commit 00568d86c0af49e00709e73d7190f502b7726389 Author: Amir Vadai Date: Tue Mar 8 18:15:59 2011 +0200 sdp: removed unneeded backport Signed-off-by: Amir Vadai commit 9655fe088ae2ee57e8b32e329020fb6a5c503c57 Author: Amir Vadai Date: Tue Mar 8 13:49:36 2011 +0200 sdp: fix backports Signed-off-by: Amir Vadai commit 16f5aab43113bbec3a66804a57a39f3b57c59d41 Author: Amir Vadai Date: Tue Mar 8 10:25:35 2011 +0200 sdp: Limit total memory consumed by rcvbuf rcvbuf is already limited by the payload in the queue. But also need to limit total memory consumption of it, since small packets received might have a very large overhead to the payload. Signed-off-by: Amir Vadai commit 84f9af7a7b5a314027e0d7abb6029ed582956380 Author: Amir Vadai Date: Tue Mar 8 10:23:24 2011 +0200 sdp: fix "sdpprf empty after a long run" sdpprf_log_count gets to a negative value after a long run. This is only a quick fix - still might loose logs sometimes. Signed-off-by: Amir Vadai commit 4dddc02bb39d0feb9618f82e5af9f8e66777ff29 Author: Amir Vadai Date: Mon Mar 7 13:11:59 2011 +0200 sdp: make SDP_RX_SIZE a module parameter Signed-off-by: Amir Vadai commit 212eacaea83a1985550ecd2f0ef3d0430b9cce54 Author: Amir Vadai Date: Sun Mar 6 14:38:08 2011 +0200 sdp: Rollback credit limit during ZCopy transaction. This limit was added at commit 2574b53 ("Abort rx SrcAvail when out of credits") Signed-off-by: Amir Vadai commit 0e73ab4abc246719417885c3ecc52fbec1f027fb Author: Amir Vadai Date: Sun Mar 6 15:09:24 2011 +0200 sdp: get per socket memory statistics at socket's sysfs file Signed-off-by: Amir Vadai commit 780584abc16afb48468970cec53719815218e70a Author: Amir Vadai Date: Sun Mar 6 14:33:00 2011 +0200 sdp: fix a hole in rx memory limit Signed-off-by: Amir Vadai commit d3646ade754a7d39eec5276eade083517a563f25 Author: Amir Vadai Date: Sun Mar 6 14:31:27 2011 +0200 sdp: make sure memory is reclaimed Signed-off-by: Amir Vadai commit af46b8597cc9e4589abbc11dea6d891189d2a3ba Author: Amir Vadai Date: Sun Mar 6 14:27:57 2011 +0200 sdp: send packets without payload when credits=1 This according to the SPEC, and prevent a deadlock in ZCopy. SrcAvailCancel wasn't acked when credits got low. Signed-off-by: Amir Vadai commit 83dd284cdc2510d5bb45d5fde7d8033c9cd52313 Author: Amir Vadai Date: Sun Mar 6 14:21:14 2011 +0200 sdp: removed some prints to sdpprf Signed-off-by: Amir Vadai commit 104d9d250fe4bbae217ab174448417cebef60df9 Author: Amir Vadai Date: Sun Mar 6 14:19:26 2011 +0200 sdp: remove unused rcvbuf_scale module parameter Was used when in slow start mechanism. Not needed till having slow start again. Signed-off-by: Amir Vadai commit 08368100e254d01bb19fdacf4b9f07ea0f258c11 Author: Amir Vadai Date: Mon Feb 28 11:57:18 2011 +0200 sdp: fix memory socket accounting skb->truesize - total bytes allocated by skb, including fragments Specific socket accounting: * sk->sk_wmem_queued - send bytes currently in TX queue * RX queue accounting is done by using seq * sk->sk_rmem_alloc - bytes consumed by RX Protocol accounting: * sk->sk_forward_alloc - bytes that are available to be consumed * prot->memory_allocated - bytes consumed by TX/RX Signed-off-by: Amir Vadai commit 329c1e5b4b4ea4b14cd31aeb8462ab2216f7de54 Author: Amir Vadai Date: Mon Feb 28 11:33:58 2011 +0200 sdp: fix sdp_sendmsg counters in sdpstats Signed-off-by: Amir Vadai commit 071041de8ba8d9ee34f7d9e50fc1d79dde707824 Author: Jeff Becker Date: Mon Sep 12 21:07:48 2011 +0300 Remove carriage returns from DOS formatted patches so they can apply Signed-off-by: Jeff Becker commit e688148a96b3b54898ccf9aa0cf192e0133f90f2 Author: Vladimir Sokolovsky Date: Mon Sep 12 19:54:26 2011 +0300 ofa_kernel.spec: Fix %files macro Signed-off-by: Vladimir Sokolovsky commit fb272719c859197931289ad12d311b639d3a0d02 Author: Jeff Becker Date: Tue Sep 6 12:42:58 2011 -0700 NFSRDMA: fixes for 2.6.32 Signed-off-by: Jeff Becker commit 09518cdc0eb04ca9a9adeb538c114eb8211ad543 Author: Mike Marciniszyn Date: Wed Sep 7 15:14:44 2011 -0400 IB/qib: Change driver default for wc_pat to 2 to work around any inconsistency of the PAT1 configuration across all CPUs. The fix for this problem has been committed by Suresh Siddha of Intel to 2.6.38.3 (https://lkml.org/lkml/2011/4/13/232) but there could be kernels without the patch backported. This patch works around this bug without any intervention by the customer or our support. Signed-off-by: Ramkrishna Vepa Signed-off-by: Mike Marciniszyn commit 6aebbfc7b7f837223715b7b667837e2434c1d46c Author: Kumar Sanghvi Date: Sun Aug 28 14:15:15 2011 +0530 iw_cxgb4: Add 2.6.16_sles10_sp4 backport Signed-off-by: Kumar Sanghvi commit 22d4ebec611954c65fd08155c0b4557e3aa983f0 Author: Vipul Pandya Date: Fri Sep 2 11:57:11 2011 +0530 cxgb4: Updates to SLES10SP4 backport Signed-off-by: Vipul Pandya commit c4bb4a6565fe15ffe6da16afeba0aae1fb6a48ab Author: Vipul Pandya Date: Fri Sep 2 11:24:44 2011 +0530 cxgb4: Add backport patches for SLES10SP4 Signed-off-by: Vipul Pandya commit 98456d970a44e7990033d4df87bbaf537d1d1866 Author: Kumar Sanghvi Date: Sun Aug 28 13:58:37 2011 +0530 backports: Add sles10_sp4 backport. Adds most of backport files from sles10_sp3 into sles10_sp4. Also, adds support in get_backport.sh script for sles10_sp4. Signed-off-by: Kumar Sanghvi commit 5790e624dd074e55cdaa680a74a1e2724bfff35e Author: Shawn Bohrer Date: Thu Aug 25 12:28:55 2011 -0500 Drop ethtool.h from the 3.0 backports ethtool_phys_id_state is already included in 3.0 so drop this patch Signed-off-by: Shawn Bohrer commit 5d249db40dc8daa233153d5c6b229b889a3129b5 Author: Jeff Becker Date: Sun Aug 28 11:11:12 2011 +0300 Remove carriage returns from DOS formatted patches. Signed-off-by: Jeff Becker commit bd205b390d800e788ce0b394d74f370681e9b3dc Merge: 7c9d10b 6128a5d Author: Vladimir Sokolovsky Date: Thu Aug 25 19:26:02 2011 +0300 Merge branch 'ofed_1_5' of git://git.openfabrics.org/~vipulp/ofed_kernel into ofed_kernel_1_5 commit 6128a5d33674df915a196f6deb3b741706b8f7c0 Author: Kumar Sanghvi Date: Sat Aug 27 12:15:49 2011 +0530 iw_cxgb4/cxgb4: Add SLES11sp1 backport Also, adds support in ofed_script/get_backport.sh for sles11sp1 Signed-off-by: Kumar Sanghvi commit 515c1fdec3c88f559abae6b41e0044b23fe6c3e9 Author: Kumar Sanghvi Date: Sun Aug 7 12:03:17 2011 +0530 iw_cxgb4: Add 2.6.30 backport Signed-off-by: Kumar Sanghvi commit be4af96779aca48bba3473be3b894e4bd11d0987 Author: Vipul Pandya Date: Thu Aug 4 18:04:00 2011 +0530 Adding patches for 2.6.30 kernel for cxgb4 driver. Signed-off-by: Vipul Pandya commit 262799f93ca7395807e2a983655c1bfc44d30e9b Author: Kumar Sanghvi Date: Mon Aug 8 12:34:22 2011 +0530 iw_cxgb4: Add 2.6.18-EL5.6 backport Signed-off-by: Kumar Sanghvi commit b428d6aec7163cc88da4a464c1b4ec596db712f8 Author: Vipul Pandya Date: Tue Aug 9 19:26:21 2011 +0530 Adding patches for RHEL5.6 for cxgb4 driver. Signed-off-by: Vipul Pandya commit 6b4bc39de10f584b1f4ff8687e90cdd308385cd2 Author: Kumar Sanghvi Date: Sun Aug 7 14:29:33 2011 +0530 iw_cxgb4: Add 2.6.18-EL5.7 backport Signed-off-by: Kumar Sanghvi commit be062eb4d43c80744db15fedf53c105e46cdd143 Author: Vipul Pandya Date: Sat Aug 6 18:38:53 2011 +0530 Adding patches for RHEL5.7 for cxgb4 driver. Signed-off-by: Vipul Pandya commit 7c9d10b0a1da56b9c4c487a59d51cf2592129383 Author: Vladimir Sokolovsky Date: Wed Aug 24 10:13:43 2011 +0300 makefile: Set the default value for INSTALL_MOD_DIR Signed-off-by: Vladimir Sokolovsky commit 5f745a059a7e14f4722bf0a7551610d2d1d2df4a Author: Vladimir Sokolovsky Date: Thu Aug 18 17:45:52 2011 +0300 Add kernel 3.0 support Signed-off-by: Vladimir Sokolovsky commit 8680bf169d1fdd9fd8ef2b0b176dc5eceb2be457 Author: Vladimir Sokolovsky Date: Thu Aug 18 18:07:01 2011 +0300 cxgb3: Add RHEL6.1 support Signed-off-by: Vladimir Sokolovsky commit ee1f65184bde2ff21ed878dc32e7364fdf29f54e Author: Vladimir Sokolovsky Date: Thu Aug 18 17:35:52 2011 +0300 get_backport_dir.sh: Add 2.6.32-71.18.2.1chaos.ch5.x86_64 kernel support Signed-off-by: Vladimir Sokolovsky commit 3d5076d0954cdbf0eb2b30cb649b21eb3dfaf84b Author: Vladimir Sokolovsky Date: Thu Aug 18 15:36:50 2011 +0300 Simplify RPM spec and makefile Signed-off-by: Vladimir Sokolovsky commit a005b31e772d294834336e14f08df6b9cfffdccd Author: Jeff Becker Date: Fri Aug 12 15:52:34 2011 -0700 NFSRDMA: remove a patch that does not apply Signed-off-by: Jeff Becker commit 21d85f665e08231a8a7c46d3e21ed6a5776f323c Author: Jeff Becker Date: Fri Aug 12 15:27:26 2011 -0700 NFSRDMA: remove patch that does not apply Signed-off-by: Jeff Becker commit ea984fa4aa3019291b5ed35e50968d9101f454c4 Author: Jeff Becker Date: Fri Aug 12 14:27:39 2011 -0700 NFSRDMA: bring patches up to date with upstream kernel Signed-off-by: Jeff Becker commit 24315d7028e58d0d457809539830eb048cc9e3fc Author: Kumar Sanghvi Date: Mon Aug 8 12:28:43 2011 +0530 iw_cxgb3: Correct Makefile to fix build errors Signed-off-by: Kumar Sanghvi commit 9efe93de87e4cf0ccabc58ebcd43ffccbe8859c3 Author: Kumar Sanghvi Date: Thu Aug 4 09:05:05 2011 +0530 iw_cxgb4: Add 2.6.31 backport Signed-off-by: Kumar Sanghvi commit f1d81fde3e21ae58cb9598a6848c899e0e59127f Author: Vipul Pandya Date: Thu Aug 4 14:27:16 2011 +0530 Adding patches for 2.6.31 kernel for cxgb4 driver. Signed-off-by: Vipul Pandya commit 612f0b519e1644301930d2cc5eb38f83e05cf114 Author: Kumar Sanghvi Date: Fri Aug 5 14:35:08 2011 +0530 iw_cxgb4 : Add 2.6.32-EL6.1 backport Signed-off-by: Kumar Sanghvi commit fd2f51115e6061798bfabd48be838f26c75d9ad0 Author: Kumar Sanghvi Date: Fri Aug 5 14:31:58 2011 +0530 cxgb4: Add 2.6.32-EL6.1 backport Also, adds support to ofed script for 2.6.32-EL6.1. Signed-off-by: Kumar Sanghvi commit e6b03350e5f88b5718c6792fe61c186ffce2b38e Author: Kumar Sanghvi Date: Thu Aug 4 06:54:43 2011 +0530 iw_cxgb4: Add 2.6.32-EL6 backport Signed-off-by: Kumar Sanghvi commit d923453a1d0cc0d1f07c4995a58ba79a923ebbba Author: Kumar Sanghvi Date: Fri Aug 5 11:21:58 2011 +0530 cxgb4: Add 2.6.32-EL6 backport Signed-off-by: Kumar Sanghvi commit facbbf761f214f7bb07bc41b33bc0e46fe38735c Author: Kumar Sanghvi Date: Wed Aug 3 06:23:25 2011 +0530 iw_cxbg4: Add 2.6.32 backport Signed-off-by: Kumar Sanghvi commit ea11038fd5c87f8682759ad6d99734547a743c87 Author: Vipul Pandya Date: Wed Aug 3 17:24:07 2011 +0530 Adding patches for 2.6.32 kernel for cxgb4 driver. Signed-off-by: Vipul Pandya commit 57b69aa666858516f745e89c7ad8ffc2c64a00c0 Author: Kumar Sanghvi Date: Mon Aug 1 02:18:38 2011 +0530 iw_cxgb4: Add 2.6.33 backport Signed-off-by: Kumar Sanghvi commit 3d19890a6692898d73ae7a2bf5f6105417ffb8a6 Author: Vipul Pandya Date: Tue Aug 2 11:33:28 2011 +0530 Adding backport patches for 2.6.33 kernel for cxgb4 driver. Signed-off-by: Vipul Pandya commit 3689cdebf7695469e5e94299c05341453e86bfe8 Author: Kumar Sanghvi Date: Sun Jul 31 13:57:43 2011 +0530 iw_cxgb4: Add 2.6.34 backport Signed-off-by: Kumar Sanghvi commit 1ca28482af8297fe2b054e2d01da096115b2c038 Author: Kumar Sanghvi Date: Wed Aug 3 03:30:48 2011 +0530 cxgb4: Updates to 2.6.34 backport Uses dev->real_num_tx_queues since netif_set_real_num_tx_queues is not available in 2.6.34. Signed-off-by: Kumar Sanghvi commit 78fe65116880406058ac28d692c20d765f41ae96 Author: Vipul Pandya Date: Mon Aug 1 18:46:16 2011 +0530 cxgb4: Add 2.6.34 backport Based on earlier backports. Backports ethtool.h, netdevice.h, skbuff.h and pci_regs.h to 2.6.34. Signed-off-by: Vipul Pandya commit 88bfe7ef260c44e78acec0bde47432a27921635e Author: Kumar Sanghvi Date: Sun Jul 31 12:06:07 2011 +0530 iw_cxgb4: Add 2.6.35 backport Signed-off-by: Kumar Sanghvi commit 9f2bfe48c13f2967b7505b383df22c85a974ca56 Author: Vipul Pandya Date: Sat Jul 30 20:40:34 2011 +0530 cxgb4: Adding 2.6.35 backport Based on earlier backports. Backports ethtool.h, netdevice.h and skbuff.h The patch also removes additional functions from ethtool_ops structure, and also removes additional functions from struct net_device_ops. Signed-off-by: Vipul Pandya commit 6232cb9bcb057a0db0d33e2be8fe3b25a007a7b3 Author: Kumar Sanghvi Date: Thu Jul 28 08:11:25 2011 +0530 iw_cxgb4: Add 2.6.36 backport Signed-off-by: Kumar Sanghvi commit f3dbb00523c16346eab4b388c3663feeebf93c14 Author: Vipul Pandya Date: Wed Jul 27 05:31:14 2011 +0530 cxgb4: Add 2.6.36 backport Includes ethtool, netdevice and sk_buff related changes Backports ethtool.h, netdevice.h and skbuff.h to 2.6.36 Signed-off-by: Vipul Pandya commit 138164e031fe6edbb0e1261eb366b07672ad7dc6 Author: Kumar Sanghvi Date: Thu Jul 28 05:43:53 2011 +0530 iw_cxgb4: Add 2.6.37 backport Signed-off-by: Kumar Sanghvi commit 9e025ca96ca12c7f4d8505ee42a2be0b63b8559b Author: Vipul Pandya Date: Tue Jul 26 08:04:49 2011 +0530 cxgb4: Add 2.6.37 backport Includes ethtool and netdevice related changes. Backports ethtool.h and netdevice.h to 2.6.37. Also, a change related to missing definition of inline function netdev_queue_numa_node_read. Signed-off-by: Vipul Pandya commit f1d5e4df3efb6a840e2c39fd238549a4b153e5e9 Author: Kumar Sanghvi Date: Thu Jul 28 03:04:58 2011 +0530 iw_cxgb4: Add 2.6.38 backport Signed-off-by: Kumar Sanghvi commit 5fdd7c5e5be5029cf35fee72d642dd5f9cb97f0b Author: Vipul Pandya Date: Thu Jul 28 02:37:49 2011 +0530 cxgb4: Add 2.6.38 backport netdevice related changes Reverts netdevice changes introduced by the mainline commit 02f8c6aee8df3cdc935e9bdd4f2d020306035dbe Signed-off-by: Vipul Pandya commit 9bb553e1f31d82a23acb78daf9dd515aa5ba45f3 Author: Vipul Pandya Date: Mon Jul 25 11:51:06 2011 +0530 cxgb4: Add 2.6.38 backport ethtool.h related changes Backport ethtool.h and related change in cxgb4 driver Signed-off-by: Vipul Pandya commit 26333b82adda24ca956df48acc7b88fb0f980e89 Author: Kumar Sanghvi Date: Mon Aug 1 09:32:35 2011 +0530 cxgb4: Corrections in 2.6.39 backport Correction to create the ethtool.h backport file directly in the kernel_addons directory. Signed-off-by: Kumar Sanghvi commit 1b41d736efa870a62fdfe3643e95913bc08d8066 Author: Jeff Becker Date: Sun Jul 31 10:38:31 2011 +0300 Remove rnfs_fs.patch from backport directories as it no longer applies Signed-off-by: Jeff Becker commit 6345bce892e00d48479b240ef31677c748b1f3b1 Merge: 74a041b e29d1be Author: Vladimir Sokolovsky Date: Thu Jul 28 18:36:15 2011 +0300 Merge branch 'ofed_1_5' of git://git.openfabrics.org/~vipulp/ofed_kernel into ofed_kernel_1_5 commit 74a041bbed802c193007f9fa3e2c75ae0424e625 Author: Mike Marciniszyn Date: Wed Jul 27 15:05:29 2011 -0400 IB/qib: Remove backport hunk from in the future Signed-off-by: Mike Marciniszyn commit 08bb16a8da0890e6c6bb8e29f809b5d3c4fcaedc Author: Mike Marciniszyn Date: Wed Jul 27 15:02:10 2011 -0400 IB/qib: Remove backport hunk from in the future Signed-off-by: Mike Marciniszyn commit e483e5e780f59b3cb9458d487f954424e59aac48 Author: Ram Vepa Date: Wed Jul 27 14:13:01 2011 -0400 IB/qib: Adjust to use cpumask_weight() and cpumask_first() The cpus_weight macro doesn't work correctly when the CONFIG_CPUMASK_OFFSTACK is chosen. The macro can walk run off the end of the valid bits in the mask because NR_CPUS is much larger than the actual number of bits. This patch adjusts the code to use the APIs that were added for kernels >= 2.6.30. Backport adjustments will need to be made for older kernels. An equivalent patch was also submitted submitted externally by KOSAKI Motohiro . Signed-off-by: Mike Marciniszyn commit f002850c98dbb0632188f551bcaae98bedc022a6 Author: Ram Vepa Date: Wed Jul 27 13:59:38 2011 -0400 IB/qib: add sysfs interface to read free contexts Indicate the number of free user contexts via the sysfs file /sys/class/infiniband/qib0/nfreectxts as required for PSM. Signed-off-by: Ram Vepa Signed-off-by: Mike Marciniszyn commit 8c9fabd5e1431143effc6cf861e4bc47f0fa0971 Author: Mitko Haralanov Date: Wed Jul 27 13:48:51 2011 -0400 IB/qib: Update active link width Update the active link width on QLE7220 chips when link goes down if chip width does not match shadowed width. Signed-off-by: Mitko Haralanov Signed-off-by: Mike Marciniszyn commit e29d1be853e7daf5a870cb363614c6cc38b8aebb Author: Kumar Sanghvi Date: Tue Jul 26 10:12:34 2011 +0530 Add 2.6.39 backport for Chelsio T4 iWarp (iw_cxgb4) driver Signed-off-by: Kumar Sanghvi commit cb0b9543e679fc1bdc17913a2bc1e4fcc6770a91 Author: Vipul Pandya Date: Thu Jul 21 10:26:36 2011 +0530 Add 2.6.39 backport for Chelsio cxgb4 driver Signed-off-by: Vipul Pandya commit 13ccee4368f6bd6b71fe3e83bb6fd0a52ba1b8b4 Author: Vipul Pandya Date: Tue Jul 19 09:12:00 2011 +0530 Add Chelsio T4 (cxgb4) and iWarp (iw_cxgb4) drivers. Also, add support in ofed scripts to build these drivers. Signed-off-by: Vipul Pandya commit 497149794a7aae2000bbac604988fbff59969463 Author: Vladimir Sokolovsky Date: Tue Jul 26 17:21:10 2011 +0300 Remove backports for unsupported OSes Signed-off-by: Vladimir Sokolovsky commit 4a10a3fb6683747f81130a8a1ebf6c976987a82d Merge: 1a1f18b a2837c4 Author: Vladimir Sokolovsky Date: Tue Jul 26 16:48:15 2011 +0300 Merge branch 'ofed_kernel_1_5_3_2' into ofed_kernel_1_5 commit 1a1f18bc88264effef6439e1703f8d063d5058bf Author: Jeff Becker Date: Tue Jul 26 16:14:00 2011 +0300 Remove SUNRPC patch that no longer applies. Signed-off-by: Jeff Becker commit bd5baae8d097a5acf475a680414cadf24d3c8233 Author: Jeff Becker Date: Tue Jul 26 16:13:05 2011 +0300 Since OFED only builds sunrpc/xprtrdma/* (instead of all of NFS), remove patches that no longer apply. Signed-off-by: Jeff Becker commit 706cc61e3b90d88555c7008371e70d8a248bbae9 Author: Jeff Becker Date: Tue Jul 26 16:08:31 2011 +0300 Patch was formatted as DOS and did not apply - convert to UNIX format Signed-off-by: Jeff Becker commit 7b50dcf01c0010fcaca5f91d80dc694295ba6b99 Author: Jeff Becker Date: Tue Jul 26 16:05:24 2011 +0300 Remove NFS patch (we don't check out NFS). Signed-off-by: Jeff Becker commit f009c232fda48636c9f321e172b0d09c11fa94c9 Author: Jeff Becker Date: Tue Jul 26 16:01:44 2011 +0300 NFSRDMA cleanup: only checkout and build xprtrdma modules Signed-off-by: Jeff Becker commit d1fe1e2ffb7e55b9b32b0f568892a4138a0d7a54 Author: Mike Marciniszyn Date: Mon Jul 25 15:33:56 2011 -0400 IB/qib: Unwind code incorrrect in qlogic_ib_init() If the init for the trace does fail the driver will attempt to destroy work queues that were never created. The destroy calls don't test for this being NULL. Reviewed-by: Ram Vepa Reviewed-by: Gary Leshner Signed-off-by: Mike Marciniszyn commit 5f675786d95583277d58d691464eede03c7ae8f0 Author: Ram Vepa Date: Mon Jul 25 12:49:09 2011 -0400 IB/qib: Fix potential deadlock with link down interrupt There is a possibility of a deadlock due to the way locks are acquired and released in qib_set_uevent_bits(). The function qib_set_uevent_bits() is called in process context calling spin_lock() and spin_unlock() while this same lock is acquired/released in interrupt context which can lead to a deadlock when running on the same cpu. The fix is to replace spin_lock() and spin_unlock() with spin_lock_irqsave() and spin_unlock_irqrestore() respectively in qib_set_uevent_bits(). Signed-off-by: Ram Vepa Signed-off-by: Mike Marciniszyn commit 218de1f4fb977f716df805d9718c4f4c49d99afa Author: Mitko Haralanov Date: Mon Jul 25 12:29:44 2011 -0400 IB/qib: Ensure that LOS and DFE are being turned off Due to timing, it is possible for the LOS and DFE to remain on. This is due to the link progressing to LinkUP prior to the driver getting the first Status Changed interrupt. By expanding the conditions under which LOS is turned off and DFE timeout is being set, timing is no longer an issue. Signed-off-by: Mitko Haralanov Signed-off-by: Mike Marciniszyn commit 0c1ce2603d0913620ab3f277dc844594405d6e63 Author: Mike Marciniszyn Date: Mon Jul 25 12:00:50 2011 -0400 IB/qib: defer hca error events to tasklet With ib_qib options as follows: options ib_qib krcvqs=1 pcie_caps=0x51 rcvhdrcnt=4096 singleport=1 ibmtu=4 A run of ib_write_bw -a yields the following: ------------------------------------------------------------------ #bytes #iterations BW peak[MB/sec] BW average[MB/sec] 1048576 5000 2910.64 229.80 ------------------------------------------------------------------ The top cpu use in a profile is: CPU: Intel Architectural Perfmon, speed 2400.15 MHz (estimated) Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (No unit mask) count 1002300 Counted LLC_MISSES events (Last level cache demand requests from this core that missed the LLC) with a unit mask of 0x41 (No unit mask) count 10000 samples % samples % app name symbol name 15237 29.2642 964 17.1195 ib_qib.ko qib_7322intr 12320 23.6618 1040 18.4692 ib_qib.ko handle_7322_errors 4106 7.8860 0 0 vmlinux vsnprintf Analysis of the stats, profile, the code, and the annotated profile indicate: - All of the overflow interrupts (one per packet overflow) are serviced on CPU0 with no mitigation on the frequency. - All of the receive interrupts are being serviced by CPU0. (That is the way truescale.cmds statically allocates the kctx IRQs to CPU.) - The code is spending all of its time servicing QIB_I_C_ERROR RcvEgrFullErr interrupts on CPU0, starving the packet receive processing - The decode_err routine is very inefficient, using a printf variant to format a "%s" and continues to loop when the errs mask has been cleared - Both qib_7322intr and handle_7322_errors read pci registers, which is very inefficient. The fix does the following: - Adds a tasklet to service the QIB_I_C_ERROR - replaces the very inefficient scnprintf() with a memcpy(). A field is added to qib_hwerror_msgs to save the sizeof("string") at compile time so that a strlen is not needed during err_decode(). - The most frequent errors (Overflows) are serviced first to exit the loop as early as possible - The loop now exits as soon as the errs mask is clear rather than fruitlessly looping through the msp array With this fix the performance changes to: ------------------------------------------------------------------ #bytes #iterations BW peak[MB/sec] BW average[MB/sec] 1048576 5000 2990.64 2941.35 ------------------------------------------------------------------ During testing of the error handling overflow patch, it was determined that some CPU's were slower when servicing both overflow and receive interrupts on CPU0 with different MSI interrupt vectors. This patch adds an option (krcvq01_no_msi) to not use a dedicated MSI interrupt for kctx's < 2 and to service them on the default interrupt. For some CPUs, the cost of the interrupt enter/exit is more costly than then the additional PCI read in the default handler. Signed-off-by: Mike Marciniszyn commit 3ff7b432701918ba887ab3970334e51c52b2983e Author: Mitko Haralanov Date: Mon Jul 25 10:52:34 2011 -0400 IB/qib: Prevent driver hang with unprogrammed boards Signed-off-by: Mitko Haralanov Signed-off-by: Mike Marciniszyn commit 7341011f3cee528b51f91050006a940abaeedd1f Author: Vladimir Sokolovsky Date: Thu Jul 7 14:41:45 2011 +0300 ipoib: Fix malformed patch ipoib_0610_set_dev_id.patch Signed-off-by: Vladimir Sokolovsky commit d6f7b51698999fac29ffd16e193f3507b2d3df35 Author: Or Gerlitz Date: Wed Jun 29 12:03:08 2011 +0300 add upstream change log to some ofed ipoib patches (fwd) ---------- Forwarded message ---------- Date: Wed, 29 Jun 2011 11:51:19 +0300 (IDT) From: Or Gerlitz To: Vladimir Sokolovsky Cc: Tziporet Koren Subject: [PATCH] add upstream change log to some ofed ipoib patches Signed-off-by: Or Gerlitz commit 156898a0d65cb4c4a020750a80f19881411b3410 Author: Dotan Barak Date: Thu Jun 23 17:58:42 2011 +0300 IB/ipoib: Add support to TSO status query This will allow ethtool to query the status of TSO and will prevent the following error message when executing 'ethtool -k ': Cannot get device tcp segmentation offload settings: Operation not supported Signed-off-by: Dotan Barak commit 9d4a0bdf8098973f4a1665c8083028cd983eaedf Author: Vladimir Sokolovsky Date: Sun May 22 11:43:24 2011 +0300 backports: Add 2.6.24.7-149 RT kernel support Signed-off-by: Vladimir Sokolovsky commit 72d7875b68f722a779bb97cba4c8b9c71b8831d4 Author: Yevgeny Petrilin Date: Wed May 4 16:47:21 2011 +0300 mlx4: Sensing port only when supported by HW Signed-off-by: Yevgeny Petrilin commit 76c0889ade3dcb94d4a2869b8ea640cb7ba35fb2 Author: Vladimir Sokolovsky Date: Sun May 1 18:14:34 2011 +0300 openibd: drop stderror from bonding's workaround Signed-off-by: Vladimir Sokolovsky commit 303d19aa19365a999a4191855044bddc765cacc6 Author: Vladimir Sokolovsky Date: Sun May 1 16:04:51 2011 +0300 openibd: drop IPoIB interfaces' configuration warnings On RHEL6.0 'ifup ib' ifconfig prints the following messages: Ifconfig uses the ioctl access method to get the full address information, which limits hardware addresses to 8 bytes. Because Infiniband address has 20 bytes, only the first 8 bytes are displayed correctly. Ifconfig is obsolete! For replacement check ip. This commit drops it to /dev/null Signed-off-by: Vladimir Sokolovsky commit dabb2e19fb30e969f4de3768143614c79aca02f3 Author: Moni Shoua Date: Sun May 1 11:52:33 2011 +0300 openibd: Add workaround for bonding over IPoIB issue BUG: http://bugs.openfabrics.org/bugzilla/show_bug.cgi?id=2259 Signed-off-by: Moni Shoua commit 04226d645a6f0e8259414fd4575463bb166fcfff Author: Christoph Lameter Date: Thu Apr 7 16:17:32 2011 +0300 IPoIB: Set pkt_type correctly for multicast packets (fix IGMP breakage) IGMP processing is broken because the IPOIB does not set the skb->pkt_type the right way for multicast traffic. All incoming packets are set to PACKET_HOST which means that igmp_recv() will ignore the IGMP broadcasts/multicasts. This in turn means that the IGMP timers are firing and are sending information about multicast subscriptions unnecessarily. In a large private network this can cause traffic spikes. Signed-off-by: Christoph Lameter Signed-off-by: Roland Dreier commit a711966cb3f7c0874bd93caf3f93dc5d1e16f79f Author: Or Gerlitz Date: Thu Apr 7 15:51:13 2011 +0300 IPoIB: Allow disabling/enabling TSO on the fly through ethtool Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier commit 0c42d37d6f2a873353eecb037b9c3298c1f271d6 Author: Or Gerlitz Date: Thu Apr 7 15:07:16 2011 +0300 IPoIB: Remove TX moderation settings from ethtool support As of commit f56bcd8 ("IPoIB: Use separate CQ for UD send completions"), there are no TX interrupts. Change the ethtool code not to report TX moderation settings, so users will not be misled to think they can control TX interrupt moderation. Pointed out by Alex Vainman Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier commit ea307a2d2441f51037e99de080a7b987479ed1fa Author: Yevgeny Petrilin Date: Thu Apr 7 16:13:56 2011 +0300 mlx4: Debug messages when cannot perform SENSE_PORT Signed-off-by: Yevgeny Petrilin commit f7dfaad0e266afc16620b72a07b7fa134033d733 Author: Yevgeny Petrilin Date: Mon Apr 4 11:15:30 2011 +0300 mlx4: Sensing port at initialization Doing this to bring the corfrect port type from start and avoid port change later. Signed-off-by: Yevgeny Petrilin commit d75f0c32a08d157fab7395e6573d46227d8a96ed Author: Yevgeny Petrilin Date: Mon Mar 28 14:35:12 2011 +0200 mlx4: Consider reserved_cqs When creting an EQ, we should reduce number of reserved cqs from the number of entries we create. Otherwise the buffer size is twice the size it should be. Signed-off-by: Yevgeny Petrilin commit 148545b7d01dc1ef6833fb6eb3077edc964b5809 Author: Nir Muchtar Date: Sun Mar 27 17:01:15 2011 +0200 mlx4_en: Fix high priority attach The problem occurs on certain conditions, when a high priority attach is the first attach on its hash bucket, it results in a low priority attach instead of the requested high priority attach. Signed-off-by: Nir Muchtar commit 787265508b5d83c7b3fb124d3c7d9546ff702d0a Author: Roel Kluin Date: Sun Mar 27 10:16:53 2011 +0200 IB/uverbs: Fix return of PTR_ERR() of wrong pointer in ib_uverbs_get_context() Signed-off-by: Roel Kluin Signed-off-by: Roland Dreier